A complete understanding of biological function at the molecular level requires an understanding both of static structure, and of the changes in structure that occur during processes such as enzyme catalysis, ligand binding and release, photocycling, and protein unfolding/refolding reactions. These changes in structure typically occur very rapidly under physiological conditions; the lifetimes of functionally crucial, structural intermediates typically lie in the range from seconds to microseconds. Such short-lived, intermediate structures are completely inaccessible to conventional x-ray crystallographic approaches. With the advent of very intense, polychromatic synchrotron x-ray sources, static x-ray diffraction patterns can be recorded from single crystals of strongly scattering proteins in the one to hundreds of milliseconds range, using a sensitive Kodak storage phosphor area detector. Advances now being made in the x-ray source itself, and in focussing optics, will reduce these exposure times by one to two orders of magnitude into the tens of microseconds to one millisecond range. If the detector is moved in its plane during the exposure, the Laue diffraction spots become streaks; and if the structure of the molecules in the crystal lattice changes during the exposure, then the intensities of the Laue diffraction streaks change, and constitute the raw data of a time-resolved x-ray crystallographic experiment. This experiment has three components, reaction initiation, in which a structural perturbation is induced, reaction monitoring and data acquisition, in which the x-ray intensities are measured as a function of time after reaction initiation, and data analysis, in which the time-dependent changes in molecular structure are identified. We propose to apply time-resolved crystallography to three main systems: the photocycle of photoactive yellow protein that appears to be a simple bacterial photosensor; the photolysis of carboxymyoglobin at low temperature; and the early stages of protein unfolding in the crystal lattice, induced by a temperature jump. The laser techniques necessary for reaction initiation and optical monitoring of the crystals will be further developed. Data analysis strategies aimed at revealing the structures of individual intermediates, rather than just their time-dependent mixture, will be explored.